K

Keyphrase Extraction Kbir Inspec

Developed by ml6team
A keyphrase extraction model fine-tuned on the Inspec dataset based on the KBIR pre-trained model, using sequence labeling to identify key phrases in text.
Downloads 22.12k
Release Time : 3/29/2022

Model Overview

This model formulates keyphrase extraction as a token classification problem through Transformer architecture, capable of accurately extracting key terms from English scientific paper abstracts.

Model Features

Multi-task Pre-training Framework
Joint optimization combining Masked Language Modeling (MLM), Keyphrase Boundary Infilling (KBI), and Keyphrase Replacement Classification (KRC)
Sequence Labeling Method
Transforms keyphrase extraction into BIO tag sequence prediction, capturing boundary information of key phrases
Domain Specialization
Fine-tuned on the Inspec dataset in computer science, suitable for academic text analysis

Model Capabilities

English keyphrase extraction
Academic text semantic analysis
Long-range context dependency capture

Use Cases

Academic Research
Automated Indexing of Paper Abstracts
Automatically extracts core terms from scientific paper abstracts, replacing manual indexing
F1@M reaches 0.564, significantly more efficient than traditional methods
Information Retrieval
Academic Literature Index Construction
Generates standardized keyword indexes for literature databases
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase